30 Years of HPC: many hardware advances, little adoption of new languages

Last summer, I had the opportunity to give the keynote at HIPS 2025—the 30th International Workshop on High-Level Parallel Programming Models and Supportive Environments. This was quite an honor since, over its history, HIPS has been a key workshop for projects like Chapel that strive to create innovative approaches in the area of emerging programming models for large-scale parallel systems and many-core architectures. To commemorate the 30th instance of HIPS, I took the approach of using my talk to reflect on the past 30 years of programming within the field of HPC, or High-Performance Computing.

Looking at the TOP500 results from 30 years ago—November 1995—we see that systems from Fujitsu, Intel, and Cray make up the top five. Core counts ranged from 80 to 3,680, and performance as measured by Rmax values ranged from 98.9 to 170 GFlop/s. Jumping forward to the latest TOP500 list from November 2025, we see systems from HPE Cray, Eviden/Bull, and Microsoft. Core counts have jumped to the millions (2,073,600–11,340,000 cores), and Rmax values range from 561 to 1809 PFlop/s. Summarizing the changes, core counts have increased by a factor of 100s to 100s of thousands, while performance has improved by factors of millions—a massive improvement driven by vector instructions, multicore CPUs, chiplets, and GPUs.

However, the dominant HPC programming notations over this same time period have remained disappointingly similar. In 1995, the dominant languages were Fortran, C, and C++. Today, they still dominate. While MPI and SHMEM remain mainstays for distributed memory, OpenMP has ruled shared-memory programming since its 1.0 spec in 1997. The biggest change has been the advent of GPUs, leading to a plethora of extensions like CUDA, HIP, and SYCL. While HPC hardware has become far more capable, we have failed to broadly adopt any new compiled programming languages.